检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Hui Wang Yan Song Zeng-Xi Li Ian McLoughlin Li-Rong Dai National Engineering Laboratory for Speech and Language Information Processing University of Science and Technology of China Hefei China Microsoft China Beijing China School of Computing University of Kent Medway UK

ISBN: (数字)9781509066315

ISBN: (纸本)9781509066322

Despite the significant progress of deep learning based speech separation methods, it remains challenging to extract and track the speech from target speakers, especially in a single-channel multiple speaker situation. Previously, the authors proposed a source-aware context network to exploit the temporal context in mixtures and estimated sources for online speech separation. In this paper, we propose a speaker-aware approach based on the source-aware context network structure, in which the speaker information is explicitly modeled by an auxiliary speaker identification branch. Then speech separation and speaker tracking can be jointly optimized by multi-task learning. Furthermore, we study the effectiveness of time-domain representation by proposing a raw sparse waveform encoder to preserve discriminative information. Experimental results on the WSJ0-2mix benchmark show that the proposed system significantly improves Signal-to-Distortion Ratio (SDR) performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Evaluation of Scholar’s Contribution to Team Based on Weighted Co-author Network 5th

Evaluation of Scholar’s Contribution to Team Based on Weigh...

引用

5th International Conference of Pioneer Computer Scientists, Engineers and Educators, ICPCSEE 2019

作者： Zhang, Xinmeng Li, Xinguang Jiang, Shengyi Li, Xia Xie, Bolin Laboratory of Language Engineering and Computing Guangdong University of Foreign Studies GuangzhouGuangdong510006 China Non-universal Language Intelligent Processing Laboratory Guangdong University of Foreign Studies GuangzhouGuangdong510006 China School of Information Science and Technology Guangdong University of Foreign Studies GuangzhouGuangdong510006 China

ISBN: (纸本)9789811501173

The contributions of scientific researchers include personal influence and talent training achievements. In this paper, using 9964 high-quality co-author scientific papers in English teaching research from China citation database from 1997 to 2016, a weighted coauthor network with variety factors is constructed. A model was proposed to calculate the author’s contribution to the research team by combining personal and network characteristics. The results reveal a variety of characteristics of the co-author networks in English teaching research field, including statistical properties, community features, and authors’ contribution to teams in this discipline. © Springer Nature Singapore Pte Ltd. 2019.

关键词： Social networking (online)

来源：评论

学校读者我要写书评

暂无评论

Research on Cognitive Diagnostic Model Based on BP Neural Network

Research on Cognitive Diagnostic Model Based on BP Neural Ne...

引用

International Conference on Measurement, Information and Control (MIC)

作者： Jin Chen Jianghao Lin XinGuang Li Laboratory for Language Engineering and Computing Guangdong University of Foreign Studies Guangzhou China

ISBN: (数字)9781728130354

ISBN: (纸本)9781728130361

This paper proposed a cognitive diagnostic model based on BP neural network and maximum likelihood function, with an ultimate goal of realizing the cognitive diagnostic computerized adaptive test. The model was verified through the actual data from 1029 British middle school students who participated in the English reading test of the Program for International Student Assessment (PISA). The experimental results show that both the model's mode accuracy rate and the average attribute accuracy rate are relatively ideal. Moreover, the model, which is of flexibility, can be effectively applied to cognitive diagnosis in small-scale classrooms.

关键词： Adaptation models Neurons Knowledge engineering Training Biological neural networks Computational modeling

来源：评论

学校读者我要写书评

暂无评论

A Two-Stage Overlapping Community Detection Based on Structure and Node Attributes in Online Social Networks 1st

A Two-Stage Overlapping Community Detection Based on Structu...

引用

1st International Workshop on Human Brain and Artificial Intelligence, HBAI 2019, held in conjunction with the 28th International Joint Conference on Artificial Intelligence, IJCAI 2019

ISBN: (纸本)9789811513978

Traditional community detection algorithms are mainly based on network structure, while ignoring a large number of node attributes. In this paper, we propose a two-stage overlapping community detection method which combines structure and attributes(tsocd-SA). First, a set of non-overlapping communities are identified by using existing community detection methods, and community attribute summaries which represents high degree homogeneous attribute value of a community are constructed according to the attributes of the special nodes in the community. Then, we propose a similarity measure between node and community based on network structure and community attribute summary. For connector nodes which connect more than one communities, each node is divided into one or more communities based on the similarity and a specific threshold r. Experimental results in online social network datasets show that our proposed method is more effective than solely focus on structural information. © 2019, Springer Nature Singapore Pte Ltd.

关键词： Social networking (online)

来源：评论

学校读者我要写书评

暂无评论

Negative information filtering algorithm based on text content in multimedia networks

引用

International Journal of Performability engineering 2019年第11期15卷 3061-3071页

作者： Chen, Wenqing Fu, Weina College of Applied Technology China University of Labor Relations Beijing100000 China College of Computer and Information Engineering Inner Mongolia Agricultural University Hohhot010012 China Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing Hunan Normal University Changsha410081 China

In the multimedia network environment, it is necessary to effectively filter negative information in the multimedia network and enhance the ability to mine and identify valid data. This paper presents a new algorithm of negative information filtering based on text content in multimedia networks. The principal component features of negative information in multimedia networks are extracted, and matched filters are designed to filter the negative information reasonably. All text contents and negative information are normalized and sorted. They are transformed into the same text format for classification and processing, and the filtering and detection of negative information are realized. Finally, based on the semantic features of text content, the support vector machine algorithm is used to extract negative information features from data. Experimental results show that the algorithm improves the filtering accuracy and performance for negative information in multimedia networks, and it has good application value. © 2019 Totem Publisher, Inc. All rights reserved.

关键词： Information filtering

来源：评论

学校读者我要写书评

暂无评论

Co-attention Networks for Aspect-Level Sentiment Analysis 8th

Co-attention Networks for Aspect-Level Sentiment Analysis

引用

8th CCF International Conference on Natural language Processing and Chinese computing, NLPCC 2019

作者： Li, Haihui Xue, Yun Zhao, Hongya Hu, Xiaohui Peng, Sancheng School of Physics and Telecommunication Engineering South China Normal University Guangzhou510006 China Industrial Central Shenzhen Polytechnic Shenzhen518055 China Laboratory of Language Engineering and Computing Guangdong University of Foreign Studies Guangzhou510006 China

ISBN: (纸本)9783030322359

Aspect-level sentiment analysis has identified its significance in sentiment polarity classification of consumer review. For the purpose of specific target sentiment analysis, we put forward a co-attentive deep learning method in the manner of human processing. To start with, the GRUs are taken to extract the hidden states of the different word embeddings. Further, via the interactive learning of the co-attention network, the representations of the target and the context can be obtained. In addition, the attention weights are determined based on the self-attention mechanism to update the final representations. The experimental results evaluated on the SemEval 2014 and Twitter establish a strong evidence of the high accuracy. © 2019, Springer Nature Switzerland AG.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

Data prioritization aware resource allocation in internet of vehicles using multi-agent deep reinforcement learning

引用

Neural networks : the official journal of the International Neural Network Society 2025年 190卷 107671页

作者： Cong Wang Yingshan Guan Sancheng Peng Hao Chen Guorui Li School of Computer and Communication Engineering Northeastern University at Qinhuangdao Qinhuangdao 066004 Hebei China Hebei Key Laboratory of Marine Perception Network and Data Processing Qinhuangdao 066004 Hebei China. Electronic address: congw@***. School of Computer and Communication Engineering Northeastern University at Qinhuangdao Qinhuangdao 066004 Hebei China. Electronic address: ***@***. Laboratory of Language Engineering and Computing Guangdong University of Foreign Studies Guangzhou 510006 Guangdong China. Electronic address: psc346@***. College of Information Science and Engineering Ritsumeikan University Ibaraki 567-8570 Osaka Japan. Electronic address: ***@***. Hebei Key Laboratory of Marine Perception Network and Data Processing Qinhuangdao 066004 Hebei China. Electronic address: lgr@***.

Intelligent transportation systems (ITS) are facing the limitation of spectral resources and stringent real time communication requirements. How to effectively allocate system resources for maximizing the performance in Internet of Vehicles (IoV) is still a substantial challenge, particularly the priority and urgency of different types of data need to be focused. To improve the allocation spectrum resources and optimize transmission power while taking into the dynamic characteristics of vehicles and data priorities account, we design a time-series-based multi-agent deep reinforcement learning framework (NL-MAPPO for short), in this paper. First, we formulate the joint optimization problem as a multi-agent Markov decision process to ensure the minimization of transmission delays and energy consumption when the total vehicle-to-vehicle (V2V) link capacity is maximized. Here, V2V link capacity refers to the maximum achievable data rate for direct communication between vehicles, which depends on factors such as signal strength, interference, and available bandwidth. Then, we design a multi-agent resource allocation algorithm based on a shared-critic mechanism to realize the global sharing of channel information and solve the optimization problem. Finally, to improve efficiency, we also introduce a time series-based channel information extraction mechanism to capture the temporal characteristics of channel information. The simulation experiments were conducted and the results demonstrated that our proposed NL-MAPPO can demonstrate superiority in multiple metrics.

关键词： Data prioritization Internet of vehicles Mobile edge computing Multi-agent deep reinforcement learning Spectrum resource allocation

来源：评论

学校读者我要写书评

暂无评论

Improving GANs for speech enhancement

arXiv

引用

arXiv 2020年

作者： Phan, Huy McLoughlin, Ian V. Pham, Lam Chén, Oliver Y. Koch, Philipp de Vos, Maarten Mertins, Alfred School of Computing University of Kent United Kingdom National Engineering Laboratory for Speech and Language Information Processing University of Science and Technology of China China Department of Engineering Science University of Oxford United Kingdom Department of Electrical Engineering KU Leuven Belgium Institute for Signal Processing University of Lübeck Germany

Generative adversarial networks (GAN) have recently been shown to be efficient for speech enhancement. Most, if not all, existing speech enhancement GANs (SEGANs) make use of a single generator to perform one-stage enhancement mapping. In this work, we propose two novel SEGAN frameworks, iterated SEGAN (ISEGAN) and deep SEGAN (DSEGAN). In the two proposed frameworks, the GAN architectures are composed of multiple generators that are chained to accomplish multiple-stage enhancement mapping which gradually refines the noisy input signals in stage-wise fashion. On the one hand, ISEGAN's generators share their parameters to learn an iterative enhancement mapping. On the other hand, DSEGAN's generators share a common architecture but their parameters are independent;as a result, different enhancement mappings are learned at different stages of the network. We empirically demonstrate favorable results obtained by the proposed ISEGAN and DSEGAN frameworks over the vanilla SEGAN. The source code is available at http://***/pquochuy/idsegan. Copyright © 2020, The Authors. All rights reserved.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Combining transformation and classification for recognizing textual entailment 10th

Combining transformation and classification for recognizing ...

引用

10th International Symposium on Intelligence Computation and Applications, ISICA 2018

作者： Ren, Han Wan, Jing Chen, Xiaomei Laboratory of Language Engineering and Computing Guangdong University of Foreign Studies Guangzhou510420 China Center for Lexicographical Studies Guangdong University of Foreign Studies Guangzhou510420 China School of Information Science and Technology Guangdong University of Foreign Studies Guangzhou510420 China

ISBN: (纸本)9789811364723

This paper introduces an approach combining transformation and classification methods for recognizing textual entailment. In transformation model, directional and undirected inference relations are recognized, and text fragments having such relations in text are replaced by the counterparts in hypothesis. In classification model, a hybrid kernel-based approach is introduced, and three kinds of features are employed for classifying entailment. Experimental results show that the combination approach achieves a better performance in comparison with the single classification system. © Springer Nature Singapore Pte Ltd. 2019.

关键词： Text processing

来源：评论

学校读者我要写书评

暂无评论

Evolution Analysis of Information Retrieval based on co-word network

Evolution Analysis of Information Retrieval based on co-word...

引用

Electronic Information Technology and Computer engineering (EITCE), International Conference on

作者： Xinmeng Zhang Xinguang Li Shengyi Jiang Xia Li Bolin Xie Laboratory of Language Engineering and Computing Guangdong University of Foreign Studies Guangzhou Guangdong China Non-universal Language Intelligent Processing Laboratory Guangdong University of Foreign Studies Guangzhou Guangdong China

ISBN: (数字)9781728135847

ISBN: (纸本)9781728135854

With the development of natural language processing and artificial intelligence, remarkable progress has been made in the field of information retrieval In this paper, we selected papers on information retrieval from web of Science in the past ten years, and constructed a Co-word network based on the co-occurrence relationship of keywords in the same paper. By analyzing the evolution of the Co-word network community, the core evolution and the number of keyword status, we discussed the evolution of research focus in the field of information retrieval in the past ten years. The results show that natural language processing, medical information retrieval, image retrieval and social media retrieval are the focus of information retrieval research in recent years.

关键词： Natural language processing Social network services Image retrieval Biomedical imaging Ontologies Market research

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：